Fix a quadratic slowdown in the Autolink extension on parsing text containing extremely long words.
Fix emphasis parsing to properly test for punctuation on platforms that support regular expression unicode categories.
Fix an issue where dashes right after a paragraph weren't parsed as horizontal rules when setext headers are disabled.
GFM autolinks will no longer include the closing bracket of a surrounding link or image.
InlineContext.hasOpenLink can now be used to query whether there is an unclosed link or image marker before the current token.
Properly require whitespace before link titles. Parse autolinks as their own nodes
Wrap autolinks in an Autolink syntax node, rather than just URL, and exclude the wrapping angle brackets from the URL nodes.
Fix a bug that could cause blockquote markers to be attached to the wrong parent node, causing them to overlap with sibling syntax nodes.
Make sure GFM autolinking accepts URLs like test.co.uk
Fix a bug in Autolink that made it fail to accept some URLs with hyphens.
The new Autolink extension (included in the GFM extension bundle) marks some types of URLs even without angle brackets.
Fix another issue in reuse of nodes when the input has gaps.
Fix another bug in incremental parsing across input gaps.
Only parse list items as tasks when there is whitespace after the checkbox brackets. Remove an unnecessary regexp operator
Fix a crash doing an incremental parse on input ranges with gaps between them.
In the stikethrough extension, ignore opening marks with a space after and closing marks with a space before them.
Fix a crash that could occur when there were gaps in the parseable ranges right at the start of a line.
First stable version.
Fix a bug that prevented style tags from built-in extensions from being applied.
This package now attached highlighting information to its syntax tree.
It is now possible to include highlighting information when defining nodes in extensions via NodeSpec.style.
Fix a bug where GFM tables occurring directly below a paragraph weren't recognized.
The BlockContext type now has a depth property providing the amount of parent nodes, and a parentType method allowing code to inspect the type of those nodes.
Fix compatibility fallback for engines with RegExp \p support.
Fix a bug where, if there were multiple extensions passed to the editor, the wrap option got dropped from the resulting configuration.
Fix a bug where an ordered list item after a nested bullet list would get treated as part of the bullet list item.
Fix a bug that caused endLeafBlock configuration to be ignored by the parser.
The module name has changed from lezer-markdown to @lezer/markdown.
MarkdownParser now extends Parser and follows its interface.
The Markdown parser no longer has its own support for nested parsing (but can be wrapped with parseCode to get a similar effect).
The new parseCode function can be used to set up a mixed-language parser for Markdown.
Fix an issue were continued paragraph lines starting with tabs could cause the parser to create a tree with invalid node positions.
Fix a bug where an unterminated nested code block could call a nested parser with a start position beyond the end of the document.
Fix a bug where the parser could return an invalid tree when forceFinish was called during a nested parse.
parseInline has been moved to MarkdownParser so that it can also be called from an inline context.
Heading nodes now have different types based on their level.
The elt helper method can now be called with a Tree to wrap the result of a nested parse in an element.
The startNested method is now exported.
BlockParser.parse's exported type was missing an argument.
Fix a bug that would cause incorrect offsets for children nested two deep in an element passed to BlockContext.addElement.
Fix table parsing when header cells are empty.
Add an extension interface. The configure method now takes more options, allowing client code to define new syntax node types and parse logic.
Add extensions for subscript, superscript, strikethrough, tables, and task lists to the distribution.
First numbered release.